Adversarial Continual Learning

Sayna Ebrahimi, Franziska Meier, Roberto Calandra +2

Continual learning aims to learn new tasks without forgetting previously learned ones. We hypothesize that representations learned to solve each task in a sequence have a shared st…